git.chise.org Git - chise/xemacs-chise.git.1/blob - lisp/utf-2000/char-db-util.el

   1 ;;; char-db-util.el --- Character Database utility -*- coding: utf-8-er; -*-
   2
   3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
   4 ;;   2007, 2008, 2009, 2010, 2011 MORIOKA Tomohiko.
   5
   6 ;; Author: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
   7 ;; Keywords: CHISE, Character Database, ISO/IEC 10646, UCS, Unicode, MULE.
   8
   9 ;; This file is part of XEmacs CHISE.
  10
  11 ;; XEmacs CHISE is free software; you can redistribute it and/or
  12 ;; modify it under the terms of the GNU General Public License as
  13 ;; published by the Free Software Foundation; either version 2, or (at
  14 ;; your option) any later version.
  15
  16 ;; XEmacs CHISE is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with XEmacs CHISE; see the file COPYING.  If not, write to
  23 ;; the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Code:
  27
  28 (require 'chise-subr)
  29 (require 'ideograph-subr)
  30
  31 (defconst unidata-normative-category-alist
  32   '(("Lu" letter        uppercase)
  33     ("Ll" letter        lowercase)
  34     ("Lt" letter        titlecase)
  35     ("Mn" mark          non-spacing)
  36     ("Mc" mark          spacing-combining)
  37     ("Me" mark          enclosing)
  38     ("Nd" number        decimal-digit)
  39     ("Nl" number        letter)
  40     ("No" number        other)
  41     ("Zs" separator     space)
  42     ("Zl" separator     line)
  43     ("Zp" separator     paragraph)
  44     ("Cc" other         control)
  45     ("Cf" other         format)
  46     ("Cs" other         surrogate)
  47     ("Co" other         private-use)
  48     ("Cn" other         not-assigned)))
  49
  50 (defconst unidata-informative-category-alist
  51   '(("Lm" letter        modifier)
  52     ("Lo" letter        other)
  53     ("Pc" punctuation   connector)
  54     ("Pd" punctuation   dash)
  55     ("Ps" punctuation   open)
  56     ("Pe" punctuation   close)
  57     ("Pi" punctuation   initial-quote)
  58     ("Pf" punctuation   final-quote)
  59     ("Po" punctuation   other)
  60     ("Sm" symbol        math)
  61     ("Sc" symbol        currency)
  62     ("Sk" symbol        modifier)
  63     ("So" symbol        other)
  64     ))
  65
  66 (defconst shuowen-radicals
  67   [?一 ?上 ?示 ?三 ?王 ?玉 ?玨 ?气 ?士 ?丨 ; 010
  68    ?屮 ?艸 ?蓐 ?茻 ?小 ?八 ?釆 ?半 ?牛 ?犛 ; 020
  69    ?告 ?口 ?凵 ?吅 ?哭 ?走 ?止 ?癶 ?步 ?此 ; 030
  70    ?正 ?是 ?辵 ?彳 ?廴 ?㢟 ?行 ?齒 ?牙 ?足 ; 040
  71    ?疋 ?品 ?龠 ?冊 ?㗊 ?舌 ?干 ?谷 ?只 ?㕯 ; 050
  72    ?句 ?丩 ?古 ?十 ?卅 ?言 ?誩 ?音 ?䇂 ?丵 ; 060
  73    ?菐 ?𠬞 ?𠬜 ?共 ?異 ?舁 ?𦥑 ?䢅 ?爨 ?革 ; 070
  74    ?鬲 ?䰜 ?爪 ?𠃨 ?鬥 ?又 ?𠂇 ?㕜 ?支 ?𦘒 ; 080
  75    ?聿 ?畫 ?隶 ?臤 ?臣 ?殳 ?殺 ?𠘧 ?寸 ?皮 ; 090
  76    ?㼱 ?攴 ?敎 ?卜 ?用 ?爻 ?㸚 ?𥄎 ?目 ?䀠 ; 100
  77    ?眉 ?盾 ?自 ?白 ?鼻 ?皕 ?習 ?羽 ?隹 ?奞 ; 110
  78    ?萑 ?𦫳 ?苜 ?羊 ?羴 ?瞿 ?雔 ?雥 ?鳥 ?烏 ; 120
  79    ?𠦒 ?冓 ?幺 ?𢆶 ?叀 ?玄 ?予 ?放 ?𠬪 ?𣦼 ; 130
  80    ?歺 ?死 ?冎 ?骨 ?肉 ?筋 ?刀 ?刃 ?㓞 ?丰 ; 140
  81    ?耒 ?𧢲 ?竹 ?箕 ?丌 ?左 ?工 ?㠭 ?巫 ?甘 ; 150
  82    ?旨 ?曰 ?乃 ?丂 ?可 ?兮 ?号 ?亏 ?喜 ?壴 ; 160
  83    ?鼓 ?豈 ?豆 ?豊 ?豐 ?䖒 ?虍 ?虎 ?虤 ?皿 ; 170
  84    ?𠙴 ?去 ?血 ?丶 ?丹 ?青 ?井 ?皀 ?鬯 ?食 ; 180
  85    ?亼 ?會 ?倉 ?入 ?缶 ?矢 ?高 ?冂 ?𩫏 ?京 ; 190
  86    ?亯 ?𣆪 ?畗 ?㐭 ?嗇 ?來 ?麥 ?夊 ?舛 ?䑞 ; 200
  87    ?韋 ?弟 ?夂 ?久 ?桀 ?木 ?東 ?林 ?才 ?叒 ; 210
  88    ?之 ?帀 ?出 ?𣎵 ?生 ?乇 ?𠂹 ?𠌶 ?華 ?𥝌 ; 220
  89    ?稽 ?巢 ?桼 ?束 ?㯻 ?囗 ?員 ?貝 ?邑 ?𨛜 ; 230
  90    ?日 ?旦 ?倝 ?㫃 ?冥 ?晶 ?月 ?有 ?明 ?囧 ; 240
  91    ?夕 ?多 ?毌 ?𢎘 ?𣐺 ?卣 ?齊 ?朿 ?片 ?鼎 ; 250
  92    ?克 ?彔 ?禾 ?秝 ?黍 ?香 ?米 ?毇 ?臼 ?凶 ; 260
  93    ?𣎳 ?林 ?麻 ?尗 ?耑 ?韭 ?瓜 ?瓠 ?宀 ?宮 ; 270
  94    ?呂 ?穴 ?㝱 ?𤕫 ?冖 ?𠔼 ?冃 ?㒳 ?网 ?襾 ; 280
  95    ?巾 ?巿 ?帛 ?白 ?㡀 ?黹 ?人 ?𠤎 ?匕 ?从 ; 290
  96    ])
  97
  98 (defun shuowen-radical (number)
  99   (aref shuowen-radicals (1- number)))
 100
 101 (defvar char-db-file-coding-system 'utf-8-mcs-er)
 102
 103 (defvar char-db-ignored-attributes '(ideographic-products))
 104
 105 (defvar char-db-coded-charset-priority-list
 106   '(ascii
 107     control-1
 108     latin-iso8859-1
 109     latin-iso8859-2
 110     latin-iso8859-3
 111     latin-iso8859-4
 112     latin-iso8859-9
 113     latin-jisx0201
 114     cyrillic-iso8859-5
 115     greek-iso8859-7
 116     thai-tis620
 117     =jis-x0208
 118     =jis-x0208@1978
 119     =jis-x0208@1983
 120     japanese-jisx0212
 121     chinese-gb2312
 122     =jis-x0208@1990
 123     chinese-cns11643-1
 124     chinese-cns11643-2
 125     chinese-cns11643-3
 126     chinese-cns11643-4
 127     chinese-cns11643-5
 128     chinese-cns11643-6
 129     chinese-cns11643-7
 130     =jis-x0213-1
 131     =jis-x0213-1@2000
 132     =jis-x0213-1@2004
 133     =jis-x0213-2
 134     korean-ksc5601
 135     chinese-isoir165
 136     katakana-jisx0201
 137     hebrew-iso8859-8
 138     chinese-gb12345
 139     latin-viscii
 140     ethiopic-ucs
 141     =big5-cdp
 142     =gt
 143     =adobe-japan1-0
 144     =adobe-japan1-1
 145     =adobe-japan1-2
 146     =adobe-japan1-3
 147     =adobe-japan1-4
 148     =adobe-japan1-5
 149     =adobe-japan1-6
 150     =hanyo-denshi/ja
 151     =hanyo-denshi/jb
 152     =hanyo-denshi/jc
 153     =hanyo-denshi/jd
 154     =hanyo-denshi/ft
 155     =hanyo-denshi/ia
 156     =hanyo-denshi/ib
 157     =hanyo-denshi/hg
 158     =hanyo-denshi/jt
 159     =hanyo-denshi/ks
 160     =daikanwa
 161     =daikanwa@rev2
 162     =daikanwa@rev1
 163     =cbeta
 164     =gt-k
 165     ideograph-hanziku-1
 166     ideograph-hanziku-2
 167     ideograph-hanziku-3
 168     ideograph-hanziku-4
 169     ideograph-hanziku-5
 170     ideograph-hanziku-6
 171     ideograph-hanziku-7
 172     ideograph-hanziku-8
 173     ideograph-hanziku-9
 174     ideograph-hanziku-10
 175     ideograph-hanziku-11
 176     ideograph-hanziku-12
 177     =>>>jis-x0208
 178     =>>>jis-x0213-1
 179     =>>>jis-x0213-2
 180     =>>jis-x0208
 181     =>>jis-x0213-1
 182     =>>jis-x0213-1@2000
 183     =>>jis-x0213-1@2004
 184     =>>jis-x0213-2
 185     =>>jis-x0208@1978
 186     =>>hanyo-denshi/ft
 187     =>>hanyo-denshi/ks
 188     =>>gt
 189     =>>daikanwa
 190     =+>jis-x0208
 191     =+>jis-x0213-1
 192     =+>jis-x0213-2
 193     =+>jis-x0208@1978
 194     =>jis-x0208
 195     =>jis-x0208@1997
 196     =>jis-x0213-1
 197     =>jis-x0213-1@2000
 198     =>jis-x0213-1@2004
 199     =>jis-x0213-2
 200     ==>ucs@bucs
 201     =>ucs@iso
 202     =>ucs@unicode
 203     =>ucs@jis
 204     =>ucs@JP
 205     =>ucs@jis/1990
 206     =>ucs@cns
 207     =>ucs@ks
 208     =+>ucs@unicode
 209     =+>ucs@cns
 210     =+>ucs@ks
 211     =>>ucs@iso
 212     =>>ucs@unicode
 213     =>>ucs@jis
 214     =>>ucs@cns
 215     =>>>ucs@iso
 216     =>>>ucs@unicode
 217     =ucs@iso
 218     =ucs@unicode
 219     =>>big5-cdp
 220     =>>gt-k
 221     =>gt
 222     =>big5-cdp
 223     =>daikanwa
 224     =big5
 225     =big5-eten
 226     =>gt-k
 227     =zinbun-oracle
 228     =>zinbun-oracle
 229     =ruimoku-v6
 230     =>>ruimoku-v6
 231     =jef-china3
 232     =shinjigen))
 233
 234
 235 ;;; @ char-db formatters
 236 ;;;
 237
 238 (defun char-db-make-char-spec (char)
 239   (let (ret char-spec)
 240     (cond ((characterp char)
 241            (cond ((and (setq ret (encode-char char '=ucs 'defined-only))
 242                        (not (and (<= #xE000 ret)(<= ret #xF8FF))))
 243                   (setq char-spec (list (cons '=ucs ret)))
 244                   (cond ((setq ret (get-char-attribute char 'name))
 245                          (setq char-spec (cons (cons 'name ret) char-spec))
 246                          )
 247                         ((setq ret (get-char-attribute char 'name*))
 248                          (setq char-spec (cons (cons 'name* ret) char-spec))
 249                          ))
 250                   )
 251                  ((setq ret
 252                         (catch 'tag
 253                           (let ((rest char-db-coded-charset-priority-list)
 254                                 ccs)
 255                             (while rest
 256                               (setq ccs (charset-name
 257                                          (find-charset (car rest))))
 258                               (if (setq ret
 259                                         (encode-char char ccs
 260                                                      'defined-only))
 261                                   (throw 'tag (cons ccs ret)))
 262                               (setq rest (cdr rest))))))
 263                   (setq char-spec (list ret))
 264                   (dolist (ccs (delq (car ret) (charset-list)))
 265                     (if (and (or (charset-iso-final-char ccs)
 266                                  (memq ccs
 267                                        '(=daikanwa
 268                                          =daikanwa@rev2
 269                                          ;; =gt-k
 270                                          =jis-x0208@1997
 271                                          ))
 272                                  (string-match "=ucs@" (symbol-name ccs)))
 273                              (setq ccs (charset-name ccs))
 274                              (null (assq ccs char-spec))
 275                              (setq ret (encode-char char ccs 'defined-only)))
 276                         (setq char-spec (cons (cons ccs ret) char-spec))))
 277                   (if (null char-spec)
 278                       (setq char-spec (split-char char)))
 279                   (cond ((setq ret (get-char-attribute char 'name))
 280                          (setq char-spec (cons (cons 'name ret) char-spec))
 281                          )
 282                         ((setq ret (get-char-attribute char 'name*))
 283                          (setq char-spec (cons (cons 'name* ret) char-spec))
 284                          ))
 285                   )
 286                  ((setq ret (get-char-attribute
 287                              char 'ideographic-combination))
 288                   (setq char-spec
 289                         (cons (cons 'ideographic-combination ret)
 290                               char-spec))
 291                   ))
 292            char-spec)
 293           ((consp char)
 294            char))))
 295
 296 (defun char-db-insert-char-spec (char &optional readable column
 297                                       required-features)
 298   (unless column
 299     (setq column (current-column)))
 300   (let (char-spec temp-char)
 301     (setq char-spec (char-db-make-char-spec char))
 302     (unless (or (characterp char) ; char
 303                 (condition-case nil
 304                     (setq char (find-char char-spec))
 305                   (error nil)))
 306       ;; define temporary character
 307       ;;   Current implementation is dirty.
 308       (setq temp-char (define-char (cons '(ideograph-daikanwa . 0)
 309                                          char-spec)))
 310       (remove-char-attribute temp-char 'ideograph-daikanwa)
 311       (setq char temp-char))
 312     (insert-char-attributes char
 313                             readable
 314                             (union (mapcar #'car char-spec)
 315                                    required-features)
 316                             nil 'for-sub-node)
 317     (when temp-char
 318       ;; undefine temporary character
 319       ;;   Current implementation is dirty.
 320       (setq char-spec (char-attribute-alist temp-char))
 321       (while char-spec
 322         (remove-char-attribute temp-char (car (car char-spec)))
 323         (setq char-spec (cdr char-spec))))))
 324
 325 (defun char-db-insert-alist (alist &optional readable column)
 326   (unless column
 327     (setq column (current-column)))
 328   (let ((line-breaking
 329          (concat "\n" (make-string (1+ column) ?\ )))
 330         name value
 331         ret al ; cal
 332         key
 333         lbs cell rest separator)
 334     (insert "(")
 335     (while alist
 336       (setq name (car (car alist))
 337             value (cdr (car alist)))
 338       (cond ((eq name 'char)
 339              (insert "(char . ")
 340              (if (setq ret (condition-case nil
 341                                (find-char value)
 342                              (error nil)))
 343                  (progn
 344                    (setq al nil
 345                          ;; cal nil
 346                          )
 347                    (while value
 348                      (setq key (car (car value)))
 349                      ;; (if (find-charset key)
 350                      ;;     (setq cal (cons key cal))
 351                      (setq al (cons key al))
 352                      ;; )
 353                      (setq value (cdr value)))
 354                    (insert-char-attributes ret
 355                                            readable
 356                                            (or al 'none) ; cal
 357                                            nil 'for-sub-node))
 358                (insert (prin1-to-string value)))
 359              (insert ")")
 360              (insert line-breaking))
 361             ((consp value)
 362              (insert (format "(%-18s " name))
 363              (setq lbs (concat "\n" (make-string (current-column) ?\ )))
 364              (while (consp value)
 365                (setq cell (car value))
 366                (if (and (consp cell)
 367                         (consp (car cell))
 368                         (setq ret (condition-case nil
 369                                       (find-char cell)
 370                                     (error nil)))
 371                         )
 372                    (progn
 373                      (setq rest cell
 374                            al nil
 375                            ;; cal nil
 376                            )
 377                      (while rest
 378                        (setq key (car (car rest)))
 379                        ;; (if (find-charset key)
 380                        ;;     (setq cal (cons key cal))
 381                        (setq al (cons key al))
 382                        ;; )
 383                        (setq rest (cdr rest)))
 384                      (if separator
 385                          (insert lbs))
 386                      (insert-char-attributes ret
 387                                              readable
 388                                              al ; cal
 389                                              nil 'for-sub-node)
 390                      (setq separator lbs))
 391                  (if separator
 392                      (insert separator))
 393                  (insert (prin1-to-string cell))
 394                  (setq separator " "))
 395                (setq value (cdr value)))
 396              (insert ")")
 397              (insert line-breaking))
 398             (t
 399              (insert (format "(%-18s . %S)%s"
 400                              name value
 401                              line-breaking))))
 402       (setq alist (cdr alist))))
 403   (insert ")"))
 404
 405 (defun char-db-insert-char-reference (plist &optional readable column)
 406   (unless column
 407     (setq column (current-column)))
 408   (let ((line-breaking
 409          (concat "\n" (make-string (1+ column) ?\ )))
 410         (separator "")
 411         name value)
 412     (insert "(")
 413     (while plist
 414       (setq name (pop plist))
 415       (setq value (pop plist))
 416       (cond ((eq name :char)
 417              (insert separator)
 418              (insert ":char\t")
 419              (cond ((numberp value)
 420                     (setq value (decode-char '=ucs value)))
 421                    ;; ((consp value)
 422                    ;;  (setq value (or (find-char value)
 423                    ;;                  value)))
 424                    )
 425              (char-db-insert-char-spec value readable)
 426              (insert line-breaking)
 427              (setq separator ""))
 428             ((eq name :radical)
 429              (insert (format "%s%s\t%d ; %c%s"
 430                              separator
 431                              name value
 432                              (ideographic-radical value)
 433                              line-breaking))
 434              (setq separator ""))
 435             (t
 436              (insert (format "%s%s\t%S" separator name value))
 437              (setq separator line-breaking)))
 438       ))
 439   (insert ")"))
 440
 441 (defun char-db-decode-isolated-char (ccs code-point)
 442   (let (ret)
 443     (setq ret
 444           (cond ((eq ccs 'arabic-iso8859-6)
 445                  (decode-char ccs code-point))
 446                 ((and (memq ccs '(=gt-pj-1
 447                                   =gt-pj-2
 448                                   =gt-pj-3
 449                                   =gt-pj-4
 450                                   =gt-pj-5
 451                                   =gt-pj-6
 452                                   =gt-pj-7
 453                                   =gt-pj-8
 454                                   =gt-pj-9
 455                                   =gt-pj-10
 456                                   =gt-pj-11))
 457                       (setq ret (decode-char ccs code-point))
 458                       (setq ret (encode-char ret '=gt 'defined-only)))
 459                  (decode-builtin-char '=gt ret))
 460                 (t
 461                  (decode-builtin-char ccs code-point))))
 462     (cond ((and (<= 0 (char-int ret))
 463                 (<= (char-int ret) #x1F))
 464            (decode-char '=ucs (+ #x2400 (char-int ret))))
 465           ((= (char-int ret) #x7F)
 466            ?\u2421)
 467           (t ret))))
 468
 469 (defvar char-db-convert-obsolete-format t)
 470
 471 (defun char-db-insert-ccs-feature (name value line-breaking)
 472   (cond
 473    ((integerp value)
 474     (insert
 475      (format
 476       (cond
 477        ((memq name '(=shinjigen
 478                      =shinjigen@1ed
 479                      =shinjigen@rev =shinjigen/+p@rev
 480                      =daikanwa/ho))
 481         "(%-18s .  %04d)\t; %c")
 482        ((eq name '=shinjigen@1ed/24pr)
 483         "(%-18s . %04d)\t; %c")
 484        ((or (memq name '(=daikanwa =>>daikanwa =>daikanwa
 485                          =daikanwa@rev1 =daikanwa@rev2
 486                          =daikanwa/+p =daikanwa/+2p
 487                          =gt =>>>gt =>>gt =>gt
 488                          =gt-k =>>gt-k =>gt-k
 489                          =>>adobe-japan1
 490                          =cbeta =>>cbeta
 491                          =zinbun-oracle =>zinbun-oracle))
 492             (string-match "^=adobe-" (symbol-name name)))
 493         "(%-18s . %05d)\t; %c")
 494        ((memq name '(=hanyo-denshi/ks =>>hanyo-denshi/ks mojikyo))
 495         "(%-18s . %06d)\t; %c")
 496        ((>= (charset-dimension name) 2)
 497         "(%-18s . #x%04X)\t; %c")
 498        (t
 499         "(%-18s . #x%02X)\t; %c"))
 500       name
 501       (if (= (charset-iso-graphic-plane name) 1)
 502           (logior value
 503                   (cond ((= (charset-dimension name) 1)
 504                          #x80)
 505                         ((= (charset-dimension name) 2)
 506                          #x8080)
 507                         ((= (charset-dimension name) 3)
 508                          #x808080)
 509                         (t 0)))
 510         value)
 511       (char-db-decode-isolated-char name value)))
 512     (if (and (= (charset-chars name) 94)
 513              (= (charset-dimension name) 2))
 514         (insert (format " [%02d-%02d]"
 515                         (- (lsh value -8) 32)
 516                         (- (logand value 255) 32))))
 517     )
 518    (t
 519     (insert (format "(%-18s . %s)" name value))
 520     ))
 521   (insert line-breaking))
 522
 523 (defun char-db-insert-relation-feature (char name value line-breaking
 524                                              ccss readable)
 525   (insert (format "(%-18s%s " name line-breaking))
 526   (let ((lbs (concat "\n" (make-string (current-column) ?\ )))
 527         separator cell sources required-features
 528         ret)
 529     (while (consp value)
 530       (setq cell (car value))
 531       (if (integerp cell)
 532           (setq cell (decode-char '=ucs cell)))
 533       (cond
 534        ((eq name '->subsumptive)
 535         (when (or (not (some (lambda (atr)
 536                                (get-char-attribute cell atr))
 537                              char-db-ignored-attributes))
 538                   (some (lambda (ccs)
 539                           (encode-char cell ccs 'defined-only))
 540                         ccss))
 541           (if separator
 542               (insert lbs))
 543           (let ((char-db-ignored-attributes
 544                  (cons '<-subsumptive
 545                        char-db-ignored-attributes)))
 546             (insert-char-attributes cell readable nil nil 'for-sub-node))
 547           (setq separator lbs))
 548         )
 549        ((characterp cell)
 550         (setq sources
 551               (get-char-attribute
 552                char (intern (format "%s*sources" name))))
 553         (setq required-features nil)
 554         (dolist (source sources)
 555           (cond
 556            ((memq source '(JP
 557                            JP/Jouyou
 558                            shinjigen shinjigen@1ed shinjigen@rev))
 559             (setq required-features
 560                   (union required-features
 561                          '(=jis-x0208
 562                            =jis-x0208@1990
 563                            =jis-x0213-1@2000
 564                            =jis-x0213-1@2004
 565                            =jis-x0213-2
 566                            =jis-x0212
 567                            =jis-x0208@1983
 568                            =jis-x0208@1978
 569                            =shinjigen))))
 570            ((eq source 'CN)
 571             (setq required-features
 572                   (union required-features
 573                          '(=gb2312
 574                            =gb12345
 575                            =iso-ir165)))))
 576           (cond
 577            ((find-charset (setq ret (intern (format "=%s" source))))
 578             (setq required-features
 579                   (cons ret required-features)))
 580            (t (setq required-features
 581                     (cons source required-features)))))
 582         (cond ((string-match "@JP" (symbol-name name))
 583                (setq required-features
 584                      (union required-features
 585                             '(=jis-x0208
 586                               =jis-x0208@1990
 587                               =jis-x0213-1-2000
 588                               =jis-x0213-2-2000
 589                               =jis-x0212
 590                               =jis-x0208@1983
 591                               =jis-x0208@1978))))
 592               ((string-match "@CN" (symbol-name name))
 593                (setq required-features
 594                      (union required-features
 595                             '(=gb2312
 596                               =gb12345
 597                               =iso-ir165)))))
 598         (if separator
 599             (insert lbs))
 600         (if readable
 601             (insert (format "%S" cell))
 602           (char-db-insert-char-spec cell readable
 603                                     nil
 604                                     required-features))
 605         (setq separator lbs))
 606        ((consp cell)
 607         (if separator
 608             (insert lbs))
 609         (if (consp (car cell))
 610             (char-db-insert-char-spec cell readable)
 611           (char-db-insert-char-reference cell readable))
 612         (setq separator lbs))
 613        (t
 614         (if separator
 615             (insert separator))
 616         (insert (prin1-to-string cell))
 617         (setq separator " ")))
 618       (setq value (cdr value)))
 619     (insert ")")
 620     (insert line-breaking)))
 621
 622 (defun insert-char-attributes (char &optional readable attributes column
 623                                     for-sub-node)
 624   (unless column
 625     (setq column (current-column)))
 626   (let (name value ; has-long-ccs-name
 627         rest
 628         radical strokes
 629         (line-breaking
 630          (concat "\n" (make-string (1+ column) ?\ )))
 631         lbs cell separator ret
 632         key al cal
 633         dest-ccss ; sources required-features
 634         ccss)
 635     (let (atr-d)
 636       (setq attributes
 637             (sort (if attributes
 638                       (if (consp attributes)
 639                           (progn
 640                             (dolist (name attributes)
 641                               (unless (memq name char-db-ignored-attributes)
 642                                 (if (find-charset name)
 643                                     (push name ccss))
 644                                 (push name atr-d)))
 645                             atr-d))
 646                     (dolist (name (char-attribute-list))
 647                       (unless (memq name char-db-ignored-attributes)
 648                         (if (find-charset name)
 649                             (push name ccss))
 650                         (push name atr-d)))
 651                     atr-d)
 652                   #'char-attribute-name<)))
 653     (insert "(")
 654     (when (memq '<-subsumptive attributes)
 655       (when (or readable (not for-sub-node))
 656         (when (setq value (get-char-attribute char '<-subsumptive))
 657           (char-db-insert-relation-feature char '<-subsumptive value
 658                                            line-breaking
 659                                            ccss readable)))
 660       (setq attributes (delq '<-subsumptive attributes)))
 661     (when (and (memq '<-denotational attributes)
 662                (setq value (get-char-attribute char '<-denotational)))
 663       (char-db-insert-relation-feature char '<-denotational value
 664                                        line-breaking
 665                                        ccss readable)
 666       (setq attributes (delq '<-denotational attributes)))
 667     (when (and (memq 'name attributes)
 668                (setq value (get-char-attribute char 'name)))
 669       (insert (format
 670                (if (> (+ (current-column) (length value)) 48)
 671                    "(name . %S)%s"
 672                  "(name               . %S)%s")
 673                value line-breaking))
 674       (setq attributes (delq 'name attributes))
 675       )
 676     (when (and (memq 'name* attributes)
 677                (setq value (get-char-attribute char 'name*)))
 678       (insert (format
 679                (if (> (+ (current-column) (length value)) 48)
 680                    "(name* . %S)%s"
 681                  "(name*              . %S)%s")
 682                value line-breaking))
 683       (setq attributes (delq 'name* attributes))
 684       )
 685     (when (and (memq 'script attributes)
 686                (setq value (get-char-attribute char 'script)))
 687       (insert (format "(script\t\t%s)%s"
 688                       (mapconcat (function prin1-to-string)
 689                                  value " ")
 690                       line-breaking))
 691       (setq attributes (delq 'script attributes))
 692       )
 693     (dolist (name '(=>ucs =>ucs*))
 694       (when (and (memq name attributes)
 695                  (setq value (get-char-attribute char name)))
 696         (insert (format "(%-18s . #x%04X)\t; %c%s"
 697                         name value (decode-char '=ucs value)
 698                         line-breaking))
 699         (setq attributes (delq name attributes))))
 700     (dolist (name '(=>ucs@gb =>ucs@big5))
 701       (when (and (memq name attributes)
 702                  (setq value (get-char-attribute char name)))
 703         (insert (format "(%-18s . #x%04X)\t; %c%s"
 704                         name value
 705                         (decode-char (intern
 706                                       (concat "="
 707                                               (substring
 708                                                (symbol-name name) 2)))
 709                                      value)
 710                         line-breaking))
 711         (setq attributes (delq name attributes))
 712         ))
 713     ;; (dolist (name '(=>daikanwa))
 714     ;;   (when (and (memq name attributes)
 715     ;;              (setq value (get-char-attribute char name)))
 716     ;;     (insert
 717     ;;      (if (integerp value)
 718     ;;          (format "(%-18s . %05d)\t; %c%s"
 719     ;;                  name value (decode-char '=daikanwa value)
 720     ;;                  line-breaking)
 721     ;;        (format "(%-18s %s)\t; %c%s"
 722     ;;                name
 723     ;;                (mapconcat (function prin1-to-string)
 724     ;;                           value " ")
 725     ;;                (char-representative-of-daikanwa char)
 726     ;;                line-breaking)))
 727     ;;     (setq attributes (delq name attributes))))
 728     (when (and (memq 'general-category attributes)
 729                (setq value (get-char-attribute char 'general-category)))
 730       (insert (format
 731                "(general-category\t%s) ; %s%s"
 732                (mapconcat (lambda (cell)
 733                             (format "%S" cell))
 734                           value " ")
 735                (cond ((rassoc value unidata-normative-category-alist)
 736                       "Normative Category")
 737                      ((rassoc value unidata-informative-category-alist)
 738                       "Informative Category")
 739                      (t
 740                       "Unknown Category"))
 741                line-breaking))
 742       (setq attributes (delq 'general-category attributes))
 743       )
 744     (when (and (memq 'bidi-category attributes)
 745                (setq value (get-char-attribute char 'bidi-category)))
 746       (insert (format "(bidi-category\t. %S)%s"
 747                       value
 748                       line-breaking))
 749       (setq attributes (delq 'bidi-category attributes))
 750       )
 751     (unless (or (not (memq 'mirrored attributes))
 752                 (eq (setq value (get-char-attribute char 'mirrored 'empty))
 753                     'empty))
 754       (insert (format "(mirrored\t\t. %S)%s"
 755                       value
 756                       line-breaking))
 757       (setq attributes (delq 'mirrored attributes))
 758       )
 759     (cond
 760      ((and (memq 'decimal-digit-value attributes)
 761            (setq value (get-char-attribute char 'decimal-digit-value)))
 762       (insert (format "(decimal-digit-value . %S)%s"
 763                       value
 764                       line-breaking))
 765       (setq attributes (delq 'decimal-digit-value attributes))
 766       (when (and (memq 'digit-value attributes)
 767                  (setq value (get-char-attribute char 'digit-value)))
 768         (insert (format "(digit-value\t . %S)%s"
 769                         value
 770                         line-breaking))
 771         (setq attributes (delq 'digit-value attributes))
 772         )
 773       (when (and (memq 'numeric-value attributes)
 774                  (setq value (get-char-attribute char 'numeric-value)))
 775         (insert (format "(numeric-value\t . %S)%s"
 776                         value
 777                         line-breaking))
 778         (setq attributes (delq 'numeric-value attributes))
 779         )
 780       )
 781      (t
 782       (when (and (memq 'digit-value attributes)
 783                  (setq value (get-char-attribute char 'digit-value)))
 784         (insert (format "(digit-value\t. %S)%s"
 785                         value
 786                         line-breaking))
 787         (setq attributes (delq 'digit-value attributes))
 788         )
 789       (when (and (memq 'numeric-value attributes)
 790                  (setq value (get-char-attribute char 'numeric-value)))
 791         (insert (format "(numeric-value\t. %S)%s"
 792                         value
 793                         line-breaking))
 794         (setq attributes (delq 'numeric-value attributes))
 795         )))
 796     (when (and (memq 'iso-10646-comment attributes)
 797                (setq value (get-char-attribute char 'iso-10646-comment)))
 798       (insert (format "(iso-10646-comment\t. %S)%s"
 799                       value
 800                       line-breaking))
 801       (setq attributes (delq 'iso-10646-comment attributes))
 802       )
 803     (when (and (memq 'morohashi-daikanwa attributes)
 804                (setq value (get-char-attribute char 'morohashi-daikanwa)))
 805       (insert (format "(morohashi-daikanwa\t%s)%s"
 806                       (mapconcat (function prin1-to-string) value " ")
 807                       line-breaking))
 808       (setq attributes (delq 'morohashi-daikanwa attributes))
 809       )
 810     (setq radical nil
 811           strokes nil)
 812     (when (and (memq 'ideographic-radical attributes)
 813                (setq value (get-char-attribute char 'ideographic-radical)))
 814       (setq radical value)
 815       (insert (format "(ideographic-radical . %S)\t; %c%s"
 816                       radical
 817                       (ideographic-radical radical)
 818                       line-breaking))
 819       (setq attributes (delq 'ideographic-radical attributes))
 820       )
 821     (when (and (memq 'shuowen-radical attributes)
 822                (setq value (get-char-attribute char 'shuowen-radical)))
 823       (insert (format "(shuowen-radical\t. %S)\t; %c%s"
 824                       value
 825                       (shuowen-radical value)
 826                       line-breaking))
 827       (setq attributes (delq 'shuowen-radical attributes))
 828       )
 829     (let (key)
 830       (dolist (domain
 831                (append
 832                 char-db-feature-domains
 833                 (let (dest domain)
 834                   (dolist (feature (char-attribute-list))
 835                     (setq feature (symbol-name feature))
 836                     (when (string-match
 837                            "\\(radical\\|strokes\\)@\\([^@*]+\\)\\(\\*\\|$\\)"
 838                            feature)
 839                       (setq domain (intern (match-string 2 feature)))
 840                      (unless (memq domain dest)
 841                        (setq dest (cons domain dest)))))
 842                   (sort dest #'string<))))
 843         (setq key (intern (format "%s@%s" 'ideographic-radical domain)))
 844         (when (and (memq key attributes)
 845                    (setq value (get-char-attribute char key)))
 846           (setq radical value)
 847           (insert (format "(%s . %S)\t; %c%s"
 848                           key
 849                           radical
 850                           (ideographic-radical radical)
 851                           line-breaking))
 852           (setq attributes (delq key attributes))
 853           )
 854         (setq key (intern (format "%s@%s" 'ideographic-strokes domain)))
 855         (when (and (memq key attributes)
 856                    (setq value (get-char-attribute char key)))
 857           (setq strokes value)
 858           (insert (format "(%s . %S)%s"
 859                           key
 860                           strokes
 861                           line-breaking))
 862           (setq attributes (delq key attributes))
 863           )
 864         (setq key (intern (format "%s@%s" 'total-strokes domain)))
 865         (when (and (memq key attributes)
 866                    (setq value (get-char-attribute char key)))
 867           (insert (format "(%s       . %S)%s"
 868                           key
 869                           value
 870                           line-breaking))
 871           (setq attributes (delq key attributes))
 872           )
 873         (dolist (feature '(ideographic-radical
 874                            ideographic-strokes
 875                            total-strokes))
 876           (setq key (intern (format "%s@%s*sources" feature domain)))
 877           (when (and (memq key attributes)
 878                      (setq value (get-char-attribute char key)))
 879             (insert (format "(%s%s" key line-breaking))
 880             (dolist (cell value)
 881               (insert (format " %s" cell)))
 882             (insert ")")
 883             (insert line-breaking)
 884             (setq attributes (delq key attributes))
 885             ))
 886         ))
 887     (when (and (memq 'ideographic-strokes attributes)
 888                (setq value (get-char-attribute char 'ideographic-strokes)))
 889       (setq strokes value)
 890       (insert (format "(ideographic-strokes . %S)%s"
 891                       strokes
 892                       line-breaking))
 893       (setq attributes (delq 'ideographic-strokes attributes))
 894       )
 895     (when (and (memq 'kangxi-radical attributes)
 896                (setq value (get-char-attribute char 'kangxi-radical)))
 897       (unless (eq value radical)
 898         (insert (format "(kangxi-radical\t . %S)\t; %c%s"
 899                         value
 900                         (ideographic-radical value)
 901                         line-breaking))
 902         (or radical
 903             (setq radical value)))
 904       (setq attributes (delq 'kangxi-radical attributes))
 905       )
 906     (when (and (memq 'kangxi-strokes attributes)
 907                (setq value (get-char-attribute char 'kangxi-strokes)))
 908       (unless (eq value strokes)
 909         (insert (format "(kangxi-strokes\t . %S)%s"
 910                         value
 911                         line-breaking))
 912         (or strokes
 913             (setq strokes value)))
 914       (setq attributes (delq 'kangxi-strokes attributes))
 915       )
 916     (when (and (memq 'japanese-radical attributes)
 917                (setq value (get-char-attribute char 'japanese-radical)))
 918       (unless (eq value radical)
 919         (insert (format "(japanese-radical\t . %S)\t; %c%s"
 920                         value
 921                         (ideographic-radical value)
 922                         line-breaking))
 923         (or radical
 924             (setq radical value)))
 925       (setq attributes (delq 'japanese-radical attributes))
 926       )
 927     (when (and (memq 'japanese-strokes attributes)
 928                (setq value (get-char-attribute char 'japanese-strokes)))
 929       (unless (eq value strokes)
 930         (insert (format "(japanese-strokes\t . %S)%s"
 931                         value
 932                         line-breaking))
 933         (or strokes
 934             (setq strokes value)))
 935       (setq attributes (delq 'japanese-strokes attributes))
 936       )
 937     (when (and (memq 'cns-radical attributes)
 938                (setq value (get-char-attribute char 'cns-radical)))
 939       (insert (format "(cns-radical\t . %S)\t; %c%s"
 940                       value
 941                       (ideographic-radical value)
 942                       line-breaking))
 943       (setq attributes (delq 'cns-radical attributes))
 944       )
 945     (when (and (memq 'cns-strokes attributes)
 946                (setq value (get-char-attribute char 'cns-strokes)))
 947       (unless (eq value strokes)
 948         (insert (format "(cns-strokes\t . %S)%s"
 949                         value
 950                         line-breaking))
 951         (or strokes
 952             (setq strokes value)))
 953       (setq attributes (delq 'cns-strokes attributes))
 954       )
 955     (when (and (memq 'shinjigen-1-radical attributes)
 956                (setq value (get-char-attribute char 'shinjigen-1-radical)))
 957       (unless (eq value radical)
 958         (insert (format "(shinjigen-1-radical . %S)\t; %c%s"
 959                         value
 960                         (ideographic-radical value)
 961                         line-breaking))
 962         (or radical
 963             (setq radical value)))
 964       (setq attributes (delq 'shinjigen-1-radical attributes))
 965       )
 966     (when (and (memq 'ideographic- attributes)
 967                (setq value (get-char-attribute char 'ideographic-)))
 968       (insert "(ideographic-       ")
 969       (setq lbs (concat "\n" (make-string (current-column) ?\ ))
 970             separator nil)
 971       (while (consp value)
 972         (setq cell (car value))
 973         (if (integerp cell)
 974             (setq cell (decode-char '=ucs cell)))
 975         (cond ((characterp cell)
 976                (if separator
 977                    (insert lbs))
 978                (if readable
 979                    (insert (format "%S" cell))
 980                  (char-db-insert-char-spec cell readable))
 981                (setq separator lbs))
 982               ((consp cell)
 983                (if separator
 984                    (insert lbs))
 985                (if (consp (car cell))
 986                    (char-db-insert-char-spec cell readable)
 987                  (char-db-insert-char-reference cell readable))
 988                (setq separator lbs))
 989               (t
 990                (if separator
 991                    (insert separator))
 992                (insert (prin1-to-string cell))
 993                (setq separator " ")))
 994         (setq value (cdr value)))
 995       (insert ")")
 996       (insert line-breaking)
 997       (setq attributes (delq 'ideographic- attributes)))
 998     (when (and (memq 'total-strokes attributes)
 999                (setq value (get-char-attribute char 'total-strokes)))
1000       (insert (format "(total-strokes       . %S)%s"
1001                       value
1002                       line-breaking))
1003       (setq attributes (delq 'total-strokes attributes))
1004       )
1005     (when (and (memq '->ideograph attributes)
1006                (setq value (get-char-attribute char '->ideograph)))
1007       (insert (format "(->ideograph\t%s)%s"
1008                       (mapconcat (lambda (code)
1009                                    (cond ((symbolp code)
1010                                           (symbol-name code))
1011                                          ((integerp code)
1012                                           (format "#x%04X" code))
1013                                          (t
1014                                           (format "%s %S"
1015                                                   line-breaking code))))
1016                                  value " ")
1017                       line-breaking))
1018       (setq attributes (delq '->ideograph attributes))
1019       )
1020     ;; (when (and (memq '->decomposition attributes)
1021     ;;            (setq value (get-char-attribute char '->decomposition)))
1022     ;;   (insert (format "(->decomposition\t%s)%s"
1023     ;;                   (mapconcat (lambda (code)
1024     ;;                                (cond ((symbolp code)
1025     ;;                                       (symbol-name code))
1026     ;;                                      ((characterp code)
1027     ;;                                       (if readable
1028     ;;                                           (format "%S" code)
1029     ;;                                         (format "#x%04X"
1030     ;;                                                 (char-int code))
1031     ;;                                         ))
1032     ;;                                      ((integerp code)
1033     ;;                                       (format "#x%04X" code))
1034     ;;                                      (t
1035     ;;                                       (format "%s%S" line-breaking code))))
1036     ;;                              value " ")
1037     ;;                   line-breaking))
1038     ;;   (setq attributes (delq '->decomposition attributes))
1039     ;;   )
1040     (if (equal (get-char-attribute char '->titlecase)
1041                (get-char-attribute char '->uppercase))
1042         (setq attributes (delq '->titlecase attributes)))
1043     (when (and (memq '->mojikyo attributes)
1044                (setq value (get-char-attribute char '->mojikyo)))
1045       (insert (format "(->mojikyo\t\t. %06d)\t; %c%s"
1046                       value (decode-char 'mojikyo value)
1047                       line-breaking))
1048       (setq attributes (delq '->mojikyo attributes))
1049       )
1050     (when (and (memq 'hanyu-dazidian-vol attributes)
1051                (setq value (get-char-attribute char 'hanyu-dazidian-vol)))
1052       (insert (format "(hanyu-dazidian-vol  . %d)%s"
1053                       value line-breaking))
1054       (setq attributes (delq 'hanyu-dazidian-vol attributes))
1055       )
1056     (when (and (memq 'hanyu-dazidian-page attributes)
1057                (setq value (get-char-attribute char 'hanyu-dazidian-page)))
1058       (insert (format "(hanyu-dazidian-page . %d)%s"
1059                       value line-breaking))
1060       (setq attributes (delq 'hanyu-dazidian-page attributes))
1061       )
1062     (when (and (memq 'hanyu-dazidian-char attributes)
1063                (setq value (get-char-attribute char 'hanyu-dazidian-char)))
1064       (insert (format "(hanyu-dazidian-char . %d)%s"
1065                       value line-breaking))
1066       (setq attributes (delq 'hanyu-dazidian-char attributes))
1067       )
1068     (unless readable
1069       (dolist (ignored '(composition
1070                          ->denotational <-subsumptive ->ucs-unified
1071                          ->ideographic-component-forms))
1072         (setq attributes (delq ignored attributes))))
1073     (while attributes
1074       (setq name (car attributes))
1075       (unless (eq (setq value (get-char-attribute char name 'value-is-empty))
1076                   'value-is-empty)
1077         (cond ((setq ret (find-charset name))
1078                (setq name (charset-name ret))
1079                (when (not (memq name dest-ccss))
1080                  (setq dest-ccss (cons name dest-ccss))
1081                  (char-db-insert-ccs-feature name value line-breaking))
1082                )
1083               ((string-match "^=>ucs@" (symbol-name name))
1084                (insert (format "(%-18s . #x%04X)\t; %c%s"
1085                                name value (decode-char '=ucs value)
1086                                line-breaking))
1087                )
1088               ((eq name 'jisx0208-1978/4X)
1089                (insert (format "(%-18s . #x%04X)%s"
1090                                name value
1091                                line-breaking))
1092                )
1093               ((and
1094                 (not readable)
1095                 (not (eq name '->subsumptive))
1096                 (not (eq name '->uppercase))
1097                 (not (eq name '->lowercase))
1098                 (not (eq name '->titlecase))
1099                 (not (eq name '->canonical))
1100                 (not (eq name '->Bopomofo))
1101                 (not (eq name '->mistakable))
1102                 (not (eq name '->ideographic-variants))
1103                 (null (get-char-attribute
1104                        char (intern (format "%s*sources" name))))
1105                 (not (string-match "\\*sources$" (symbol-name name)))
1106                 (null (get-char-attribute
1107                        char (intern (format "%s*note" name))))
1108                 (not (string-match "\\*note$" (symbol-name name)))
1109                 (or (eq name '<-identical)
1110                     (eq name '<-uppercase)
1111                     (eq name '<-lowercase)
1112                     (eq name '<-titlecase)
1113                     (eq name '<-canonical)
1114                     (eq name '<-ideographic-variants)
1115                     ;; (eq name '<-synonyms)
1116                     (string-match "^<-synonyms" (symbol-name name))
1117                     (eq name '<-mistakable)
1118                     (when (string-match "^->" (symbol-name name))
1119                       (cond
1120                        ((string-match "^->fullwidth" (symbol-name name))
1121                         (not (and (consp value)
1122                                   (characterp (car value))
1123                                   (encode-char
1124                                    (car value) '=ucs 'defined-only)))
1125                         )
1126                        (t)))
1127                     ))
1128                )
1129               ((or (eq name 'ideographic-structure)
1130                    (eq name 'ideographic-combination)
1131                    (eq name 'ideographic-)
1132                    (eq name '=decomposition)
1133                    (char-feature-base-name= '=decomposition name)
1134                    (char-feature-base-name= '=>decomposition name)
1135                    ;; (string-match "^=>*decomposition\\(@[^*]+\\)?$"
1136                    ;;               (symbol-name name))
1137                    (string-match "^\\(->\\|<-\\)[^*]*$" (symbol-name name))
1138                    (string-match "^\\(->\\|<-\\)[^*]*\\*sources$"
1139                                  (symbol-name name))
1140                    )
1141                (char-db-insert-relation-feature char name value
1142                                                 line-breaking
1143                                                 ccss readable))
1144               ((memq name '(ideograph=
1145                             original-ideograph-of
1146                             ancient-ideograph-of
1147                             vulgar-ideograph-of
1148                             wrong-ideograph-of
1149                             ;; simplified-ideograph-of
1150                             ideographic-variants
1151                             ;; ideographic-different-form-of
1152                             ))
1153                (insert (format "(%-18s%s " name line-breaking))
1154                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1155                      separator nil)
1156                (while (consp value)
1157                  (setq cell (car value))
1158                  (if (and (consp cell)
1159                           (consp (car cell)))
1160                      (progn
1161                        (if separator
1162                            (insert lbs))
1163                        (char-db-insert-alist cell readable)
1164                        (setq separator lbs))
1165                    (if separator
1166                        (insert separator))
1167                    (insert (prin1-to-string cell))
1168                    (setq separator " "))
1169                  (setq value (cdr value)))
1170                (insert ")")
1171                (insert line-breaking))
1172               ((consp value)
1173                (insert (format "(%-18s " name))
1174                (setq lbs (concat "\n" (make-string (current-column) ?\ ))
1175                      separator nil)
1176                (while (consp value)
1177                  (setq cell (car value))
1178                  (if (and (consp cell)
1179                           (consp (car cell))
1180                           (setq ret (condition-case nil
1181                                         (find-char cell)
1182                                       (error nil))))
1183                      (progn
1184                        (setq rest cell
1185                              al nil
1186                              cal nil)
1187                        (while rest
1188                          (setq key (car (car rest)))
1189                          (if (find-charset key)
1190                              (setq cal (cons key cal))
1191                            (setq al (cons key al)))
1192                          (setq rest (cdr rest)))
1193                        (if separator
1194                            (insert lbs))
1195                        (insert-char-attributes ret
1196                                                readable
1197                                                al ; cal
1198                                                nil 'for-sub-node)
1199                        (setq separator lbs))
1200                    (setq ret (prin1-to-string cell))
1201                    (if separator
1202                        (if (< (+ (current-column)
1203                                  (length ret)
1204                                  (length separator))
1205                               76)
1206                            (insert separator)
1207                          (insert lbs)))
1208                    (insert ret)
1209                    (setq separator " "))
1210                  (setq value (cdr value)))
1211                (insert ")")
1212                (insert line-breaking))
1213               (t
1214                (insert (format "(%-18s" name))
1215                (setq ret (prin1-to-string value))
1216                (unless (< (+ (current-column)
1217                              (length ret)
1218                              3)
1219                           76)
1220                  (insert line-breaking))
1221                (insert " . " ret ")" line-breaking)
1222                ;; (insert (format "(%-18s . %S)%s"
1223                ;;                 name value
1224                ;;                 line-breaking))
1225                )
1226               ))
1227       (setq attributes (cdr attributes)))
1228     (insert ")")))
1229
1230 (defun insert-char-data (char &optional readable
1231                               attributes)
1232   (save-restriction
1233     (narrow-to-region (point)(point))
1234     (insert "(define-char
1235   '")
1236     (insert-char-attributes char readable attributes)
1237     (insert ")\n")
1238     (goto-char (point-min))
1239     (while (re-search-forward "[ \t]+$" nil t)
1240       (replace-match ""))
1241     ;; from tabify.
1242     (goto-char (point-min))
1243     (while (re-search-forward "[ \t][ \t][ \t]*" nil t)
1244       (let ((column (current-column))
1245             (indent-tabs-mode t))
1246         (delete-region (match-beginning 0) (point))
1247         (indent-to column)))
1248     (goto-char (point-max))
1249     ;; (tabify (point-min)(point-max))
1250     ))
1251
1252 (defun insert-char-data-with-variant (char &optional printable
1253                                            no-ucs-unified
1254                                            script excluded-script)
1255   (insert-char-data char printable)
1256   (let ((variants (char-variants char))
1257         rest
1258         variant vs ret)
1259     (setq variants (sort variants #'<))
1260     (setq rest variants)
1261     (setq variants (cons char variants))
1262     (while rest
1263       (setq variant (car rest))
1264       (unless (get-char-attribute variant '<-subsumptive)
1265         (if (and (or (null script)
1266                      (null (setq vs (get-char-attribute variant 'script)))
1267                      (memq script vs))
1268                  (or (null excluded-script)
1269                      (null (setq vs (get-char-attribute variant 'script)))
1270                      (not (memq excluded-script vs))))
1271             (unless (and no-ucs-unified (get-char-attribute variant '=ucs))
1272               (insert-char-data variant printable)
1273               (if (setq ret (char-variants variant))
1274                   (while ret
1275                     (or (memq (car ret) variants)
1276                         ;; (get-char-attribute (car ret) '<-subsumptive)
1277                         (setq rest (nconc rest (list (car ret)))))
1278                     (setq ret (cdr ret)))))))
1279       (setq rest (cdr rest)))))
1280
1281 (defun insert-char-range-data (min max &optional script excluded-script)
1282   (let ((code min)
1283         char)
1284     (while (<= code max)
1285       (setq char (decode-char '=ucs code))
1286       (if (encode-char char '=ucs 'defined-only)
1287           (insert-char-data-with-variant char nil 'no-ucs-unified
1288                                          script excluded-script))
1289       (setq code (1+ code)))))
1290
1291 (defun write-char-range-data-to-file (min max file
1292                                           &optional script excluded-script)
1293   (let ((coding-system-for-write char-db-file-coding-system))
1294     (with-temp-buffer
1295       (insert (format ";; -*- coding: %s -*-\n"
1296                       char-db-file-coding-system))
1297       (insert-char-range-data min max script excluded-script)
1298       (write-region (point-min)(point-max) file))))
1299
1300 (defvar what-character-original-window-configuration)
1301
1302 ;;;###autoload
1303 (defun what-char-definition (char)
1304   (interactive (list (char-after)))
1305   (let ((buf (get-buffer-create "*Character Description*"))
1306         (the-buf (current-buffer))
1307         (win-conf (current-window-configuration)))
1308     (pop-to-buffer buf)
1309     (make-local-variable 'what-character-original-window-configuration)
1310     (setq what-character-original-window-configuration win-conf)
1311     (setq buffer-read-only nil)
1312     (erase-buffer)
1313     (condition-case err
1314         (progn
1315           (insert-char-data-with-variant char 'printable)
1316           (unless (char-attribute-alist char)
1317             (insert (format ";; = %c\n"
1318                             (let* ((rest (split-char char))
1319                                    (ccs (pop rest))
1320                                    (code (pop rest)))
1321                               (while rest
1322                                 (setq code (logior (lsh code 8)
1323                                                    (pop rest))))
1324                               (decode-char ccs code)))))
1325           ;; (char-db-update-comment)
1326           (set-buffer-modified-p nil)
1327           (view-mode the-buf (lambda (buf)
1328                                (set-window-configuration
1329                                 what-character-original-window-configuration)
1330                                ))
1331           (goto-char (point-min)))
1332       (error (progn
1333                (set-window-configuration
1334                 what-character-original-window-configuration)
1335                (signal (car err) (cdr err)))))))
1336
1337
1338 ;;; @ end
1339 ;;;
1340
1341 (provide 'char-db-util)
1342
1343 ;;; char-db-util.el ends here